Type B Characterization Tasks

Mark's Section

Info we need to know/measure/document LIGO-T1100378.

Generic info (OK to analyze or measure one typical item)

Device-specific info

Enzo's Section

Here we will clarify the steps to follow to complete the characterization of the Type B suspensions:

SR3

Task

August 27-31

September 3-7

September 10-13

September 14

Reference

Comments

IM Local damping

-

-

-

-

GAS Local damping

-

-

-

-

TM Local damping

-

-

-

-

IM-TM Hierarchical damping

-

-

-

-

IP Blending filter

-

-

-

-

IP Local damping

-

-

-

-

IM Diagonalization

-

-

-

-

IP Diagonalization

-

-

-

-

Fabian's proposal

The following list is yet not a comprehensive or detailed list of tasks. So far it is still at the brainstorming level and requires further categorization, like adding short and long terms tags, removing items which are not relevant, etc. Some of tasks are already in Mark’s section above. Nevertheless, for the purposes of me understanding better I wanted to write them within a larger context.

Performance

Goals:

1. Proof the Type B system works per the requirement and write a paper afterwards.

2. Quantify performance of each suspension in case they are different (due to different noise levels in sensors and actuators, magnetic forces, etc.)

3. Propose mitigation strategies according to the results from previous item 2.

Loose items that will be useful for the goals above

1. Calculate the velocity requirement for BS and SR mirrors in order to acquire the lock.

2. Quantify unexpected effects, including ones which are not easily modelled (e.g. magnetic forces and cabling) and determine any degradation in performance. We can begin by repeating Hatoya-san’s measurement for SRM which doesn’t have the ferromagnetic connectors on top of the SF.

3. Estimate the sensitivity of the oplev.

a. Ideally we should lock the target optic and measure the readout. However, locking it properly is cumbersome once it’s inside the chamber and we might not want to disturb it. Another possibility would be to set up an oplev on an optical table and fix the target mirror rigidly. Do we have components to do this? Nevertheless, the dominant noise sources are likely the vibrations of the platforms supporting optical components and fixing the mirror might not be necessary. We just have to be able to distinguish the resonant peaks of the optic from the oplev spectrum.

b. Estimate the power reaching the QPD for each suspension and calculate the shot noise level. It seems the beam in SRM will be particularly dim.

c. Put accelerometers (TEAC 710, PEM equipment) on oplev platforms and measure the effect of vibrations with fans on and off.

d. Use microphones in order to measure the effect of acoustic noise on vibrations. It will likely be negligible with fans off.

4. Measure ground motion in the vicinity of the Type B suspensions. It seems there was a Trillium accelerometer installed close to the BS for some time. I’ll check whether they got a measurement we can use. There seems to be another one installed next to the IMC rack.

5. Measure the amplitude of the vibrations of the supporting frames. Should we install a Trillium on the top platform? We don’t have many holes on the platform but we might be able to design a support using fancy clamps. Another option would be to fix geophones on the frame as was done in the prototype. See this link.

6. Is it possible to measure the transfer function from ground vibrations to OSEM readout, LVDTs and oplev by mounting inertial shakers and an accelerometer on top of the frame? PEM group has one shaker (works from 2 Hz to 3 kHz) and I have the impression NAOJ has another one. I’ll check with Takahashi-san.

7. Identify effects of possible differences between IP legs. Takahashi-san and Arai-san can suggest some strategies because they faced the problem in TAMA300. In their case one degree of freedom of the IP would become unstable whereas others remained stable.

8. What will be the effect of ground tilt on the IPs. As far as I remember this is a well-known issue in GW detector suspensions (cite Lenz).

9. Getting acquainted with the measurement of the TF using the main interferometer. I guess the main interferometer people did this measurement in bKAGRA phase 1.

10. Transfer functions and power spectra.

  1. Resolution according to measured sensitivities of LVDTs, OSEMs and coil drivers.
  2. Elucidate the origin of the increasing value/plateau of the transfer functions at higher frequencies. This might be important if we want estimate the saturation in the GAS filter performance from using the magic want. Currently I’m considering two possible causes:
    1. Coupling from coils actuators to LVDTs and OSEM readout. This effect can be measured by mechanically locking the sensor and measuring the transfer function. The effect of filters should be taken into account. See this link, this other one and this one. See Takanori’s message written to Fabian on the 2nd of October of 2015.
    2. Coupling happening some place at the data acquisition electronics. According also to Takanori such a feature appears when measuring with the OSEMs cables completely disconnected. He told me this verbally and I remember seeing a plot but I haven’t found the plot again. This was reported in the payload prototype paper.

11. Estimate the GAS filters performance saturation at higher frequencies. We didn’t adjust the magic wands in any special way other than the 3D CAD value.

12. Glitches. How to identify glitches in feedback systems? This may be a complicated problem in a feedback systems where the effect of a glitch quickly propagates. I asked Pil-Jong-san, whose project is glitch origin identification, but he still hasn’t developed a method.

13. Diagonalization

  1. Low frequency versus higher frequency diagonalization: I have the impression the low frequency diagonalization would work at higher frequencies but not the other way around. The mechanical response of mechanical systems is larger below the resonant frequencies. b. According to klog entry 6118, the frequency for diagonalization must be selected such that the mechanical transfer function is real. However, I have the impression we can just consider the magnitude and ignore the phase content when the TF is complex. Test this idea.

14. Decide a strategy to predict performance:

  1. Calculate the transfer functions from first principles (with Takanori’s Matlab/Simulink software for example) and add the measured noise.
  2. Use measured transfer functions and add measured noise.
  3. Both above.

KAGRA/Subgroups/VIS/TypeB/Characterization (last edited 2018-12-10 12:10:08 by fabian.arellano)